Arsenic removal from drinking water using granular

ferric hydroxide
OS Thirunavukkarasu1, T Viraraghavan2* and KS Subramanian3
1

Environmental Systems Engineering, Faculty of Engineering, University of Regina, Regina, Saskatchewan, Canada S4S 0A2
2
Product Safety Bureau, Health Canada, Ottawa, Ontario, Canada K1A 0K9

Abstract
This paper examines the use of granular ferric hydroxide (GFH) to remove both arsenate [As(V)] and arsenite [As(III)] present in
drinking water by conducting batch and column studies. The kinetic studies were conducted as a function of pH, and less than 5
g/l was achieved from an initial concentration of 100 g/l for both As(III) and As(V) with GFH at a pH of 7.6, which is in the
pH range typically encountered in drinking water supplies. In the isotherm studies, the observed data fitted well with both the
Freundlich and the Langmuir models. In continuous column tests (five cycles) with tap water using GFH, consistently less than
5 g/l of arsenic was achieved in the finished water for 38 to 42 hours of column operation, where the influent had a spiked arsenic
concentration of 500 g/l. High bed volumes (1260 and 1140) up to a breakthrough concentration of 5 g/l were achieved in the
column studies. The adsorptive capacities for GFH estimated from the column studies were higher than that of activated alumina
reported in the previous studies. Speciation of a natural water sample with arsenic showed the dominance of As(III) species over
As(V). Batch and column studies showed that granular ferric hydroxide (GFH) can be effectively used in small water utilities to
achieve less than 5 g As/l in drinking water.

Keywords: adsorption, drinking water, arsenic removal, granular ferric hydroxide, arsenic speciation

Nomenclature
C
Ce
C0
k
x
m
b
Q0
K
k
n
Q
qo
V
R
r

equilibrium concentration (g/l)

effluent adsorbate concentration (g/l)
influent adsorbate concentration (g/l)
Thomas rate constant (ml/min.g)
mass of solute adsorbed (g)
mass of adsorbent (g)
a constant related to the energy or net enthalpy of
adsorption (l/g)
mass of adsorbed solute completely required to saturate
a unit mass of adsorbent (g/g)
the Freundlich constant indicative of the adsorption
capacity of the adsorbent (l/g)
rate constant for adsorption (g/hg)
experimental constant indicative of the extent of
adsorption of the adsorbent
volumetric flow rate (ml/min)
maximum solid phase concentration (g/g)
throughput volume (ml)
separation factor estimated from the Langmuir constant
regression coefficient.

Introduction
Arsenic contamination of surface and subsurface waters is reported
in many parts of the world and is considered a global issue. As a
naturally occurring toxic substance in the earths crust, arsenic

* To whom all correspondence should be addressed.

Fax: +1-306-585-4855; e-mail: t.viraraghavan@uregina.ca
Received 30 August 2001; accepted in revised form 15 November 2002.

Available on website http://www.wrc.org.za

enters into aquifers and wells through natural processes, and to the
water cycle as a result of anthropogenic activities. Arsenic contamination of subsurface waters is believed to be geological. High
arsenic concentrations may result from dissolution of, or desorption
from iron oxide, and oxidation of arsenic pyrites (Welch et al.,
1999). The severity of arsenic pollution of groundwater is reported
in Bangladesh, where most of the people rely on tube wells as a
source of drinking water. It is estimated that 30 to 70 million people
in Bangladesh are at risk due to the exposure of arsenic contaminated water (Chowdhury et al., 1999; Ward, 2000). Until recently,
occurrence of arsenic in Bangladesh water supplies was believed to
be caused by pyrite oxidation; however, recent studies showed that
the causative mechanism of arsenic release to groundwater was
reductive dissolution of As-rich Fe oxyhydroxide. The reduction
was driven by microbial degradation of organic matter, which was
present in concentrations as high as 6% C in water (Nickson et al.,
2000; McArthur et al., 2001).
Ingestion of inorganic arsenic can result in both cancer (skin,
lung and urinary bladder) and non-cancer effects (NRC, 1999). The
acute and chronic toxicity effects of the ingestion of arseniccontaminated water have been well documented. Population-based
studies showed that arsenite [As(III)] and arsenate [As(V)] may
adversely affect several organs in the human body (Tseng et al.,
1968; Smith et al., 1998; Mazumder et al., 1998; Subramanian and
Kosnett, 1998; Ma et al., 1999; Chowdhury et al., 1999; Karim,
2000). Since the majority of the people affected world-wide live in
small communities, it makes sense to develop a treatment technology tailored for small communities. Furthermore, a reduction in
acceptable consumption levels of arsenic by the regulatory agencies is forcing water utilities to identify and implement costeffective arsenic removal technologies.
Treatment technologies that have been tested to remove arsenic
from drinking water under both laboratory and pilot-scale studies
were summarised by Viraraghavan et al. (1994). Iron-based salts

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

161

have been used as coagulants for arsenic removal from drinking

water, and were found to be effective in the case of large-scale water
utilities (Scott et al., 1995; Cheng et al., 1994; Sancha, 1999).
Bench-scale studies were conducted with various iron oxides to
remove arsenic from drinking water (Pierce and Moore, 1980;
Pierce and Moore, 1982; Hsia et al., 1994; Wilkie and Hering,
1996; Joshi and Chaudhuri, 1996; Raven et al., 1998; Driehaus et
al., 1998; Viraraghavan et al., 2000; Thirunavukkarasu et al.,
2001). Recent studies (Korte and Fernando, 1991; Chada, 2000)
showed that As(III) is more prevalent in groundwater than As(V).
An effective technology is sought that would remove both As(III)
and As(V) present in drinking water.
After a careful review, the United States Environmental Protection Agency (USEPA) suggested ion exchange, activated alumina,
reverse osmosis, modified coagulation/filtration, and modified
lime softening as best available technologies for As(V) removal.
However, the importance of iron-based coagulation-assisted
microfiltration, iron oxide-coated sand, and granular ferric hydroxide (GFH) was stressed for arsenic removal. Rigorous testing is
necessary to validate these technologies (USEPA, 1999; USEPA,
2000). In addition, it is essential to study the speciation changes to
establish the occurrence and toxicity of species present in the
drinking water. The objective of the present study was to assess the
potential and applicability of granular ferric hydroxide (GFH) for
removing both As(III) and As(V) present in drinking water. Batch
kinetic and isotherm studies were conducted to study the extent of
arsenic adsorption onto GFH. Column studies were conducted at
normal filtration rates to study the removal of arsenic species that
were added to the tap water. In studies with natural water, a
speciation technique was used to observe the speciation changes in
the treated water.

TABLE 1
Water quality parameters
Parameters

pH
Iron
Manganese
Turbidity (NTU)
Chloride
Copper
Zinc
Lead
Cadmium
Barium
Chromium
Chlorine (residual)
TOC

Tap water*

Kelliher water*

7.6
0.07
0.02
0.28
12
0.001
<0.005
<0.002
<0.001
0.073
<0.001
0.2
1.6

7.4
2.1
1.2
Not analysed
Not analysed
0.04
0.01
0.002
<0.001
0.011
0.001
Not analysed
Not analysed

* all parameters except pH and turbidity are in mg/l

turing process, GFH was produced from a ferric chloride solution

by neutralisation and precipitation with sodium hydroxide. The
ferric hydroxide precipitate was centrifuged and granulated by a
high-pressure process. The GFH consists of ferric oxihydroxide
(approximately 52 to 57% by mass), 43 to 48% by mass moisture
content and grain porosity of 72 to 77% (Driehaus et al., 1998).
The grain size of the GFH obtained from the manufacturer ranged
between 0.32 and 2 mm. The GFH was sieved to a size of 0.8 to
1.2 mm and used in the studies.

Materials and experimental methods

Batch studies
Water and standards
Natural water (containing arsenic) from Kelliher Water Treatment
Plant, Kelliher, Saskatchewan, Canada and tap water from the City
of Regina, Saskatchewan, Canada were used in the batch and
column studies. The major physicochemical characteristics of the
Regina tap water and Kelliher natural water are listed in Table 1.
Distilled (double) deionised water was used in the preparation of
standard solutions and for dilution purposes. As(V) stock solution
(1 000 mg/l) was prepared by dissolving 4.164 g of sodium
hydrogen arsenate heptahydrate (Na2HAsO4. 7H2O; 99.7% purity;
Sigma Chemical, Ontario) in 1 l distilled water and was preserved
with 0.5% trace metal grade HNO3 (Fisher Scientific, Ontario).
Fresh stock solution was prepared once in 30 d. One mg/l of As (V)
was prepared by pipetting 1 ml of stock solution into a 1 l
volumetric flask, and making up the solution to 1 l with distilled
water. One mg/l of As(III) stock solution was prepared by pipetting
1 ml of arsenic oxide (1 000 mg/l reference solution; Fisher
Scientific, Ontario) into a 1 l volumetric flask, and making up the
solution to 1 l with distilled water. In both the cases [As(V) or
As(III)] required working standards were prepared daily from the
stock solution. All glassware and sample bottles were washed with
a detergent solution, rinsed with tap water, soaked with 10% nitric
acid for at least 12 h, and finally rinsed with distilled water three
times.
Granular ferric hydroxide (GFH)
Granular ferric hydroxide (GFH) was obtained from the manufacturer Wasserchemie GmbH and Co.KG, Germany. In the manufac-

162

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

Batch kinetic studies were conducted as a function of pH to study

the removal of both As(III) and As(V), spiked to required concentrations in tap water. In the isotherm studies, arsenic [As(III) and
As(V)] removal was studied at the normal pH (7.6) of the tap water
in Regina. Isotherm studies were also conducted with raw water
collected from the Kelliher Water Treatment Plant, Kelliher, Saskatchewan, Canada. The raw water (containing arsenic) was collected using 18.9 l (5 US gallon) low-density polyethylene containers and studies were conducted immediately on receipt of samples
in the laboratory. In both kinetic and isotherm studies, 100 ml
samples were transferred to 250 ml Erlenmeyer flasks containing
GFH and the flasks were sealed with parafilm. The samples were
placed on a mechanical shaker and shaken at 175 r/min. All
experiments were conducted at the room temperature of 22 10C.
A portable bench top platform shaker (New Brunswick Scientific
Co. Inc., NJ, USA) was used to conduct the batch studies.
The mass of GFH used was kept at 0.2 g in the kinetic studies.
The initial concentration of both As(III) and As(V) in the tap water
was 100 g/l. The effect of solution pH on adsorption of arsenic
species on to GFH was studied at the pH levels of 5, 6, 7.6, and 8.5.
Isotherm studies were conducted by varying the mass of GFH. The
initial concentrations of As(V) and As(III) were same as in kinetic
studies. The latter studies were conducted only at pH 7.6. The
equilibrium time (6 h) from the kinetic studies was kept as contact
time. Samples were collected at regular time intervals and analysed
for residual As through graphite furnace atomic absorption
spectrometry (GFAAS). In the case of natural water samples,
soluble As, As(III) and As(V) in the samples were determined as
per speciation protocol.

Available on website http://www.wrc.org.za

Column studies

n is experimental constant indicative of the adsorption intensity

of the adsorbent.

In the column studies, all the tests were conducted in the downflow
mode and at the normal pH (7.6) of the tap water in Regina, Canada.
Both As(III) and As(V) removals were investigated. Water was
pumped through the packed column with a peristaltic pump (Model
# 7553-70, Cole Parmer Instrument Company, Ontario). The
packed volume of GFH in the column (16 mm diameter C 400 mm
long) was 43 ml, ensuring enough headspace to allow for expansion
of the medium during backwashing. The flow rate was kept
constant at 21.5 ml/min (5 m/h or 2 gpm/ft2), which yielded an
empty bed contact time (EBCT) of 2 min. A column test was also
conducted using GFH to remove arsenic from the natural water. In
this study, the volume of GFH, EBCT and flow rate were 30.5 ml,
1.42 min and 21.5 ml/min (5 m/h or 2 gpm/ft2), respectively.
In the column studies, the initial As(V) and As(III) concentrations were kept at 500 g/l in the tap water, whilst the experiments
were conducted at pH 7.6. Five cycles of downflow column tests
were conducted to evaluate the performance of GFH related to
As(V) and As(III) removal from the tap water. The first cycle was
conducted up to exhaustion for both As(III) and As(V) and for the
subsequent cycles the column test was conducted until the effluent
arsenic level reached 100 g/l. Before commencement of each
cycle, the column was backwashed with deionised water until the
effluent level reached a level that was less than 10 g/l. Samples
from the column tests were collected at regular time intervals and
analysed for residual As through GFAAS. In the column test using
natural water, the effluent samples were speciated to find soluble
As, As(III) and As(V) using the speciation protocol. The speciation
protocol used in this study was similar to that of Thirunavukkarasu
et al. (2001). The only difference was the use of anion-exchange
resin (Dowex 1X8 - 50, Sigma Chemical Co., MO, USA) of 20 to
50 mesh size in the speciation protocol instead of 50 to 100 mesh
size.
Data analysis
The data obtained from the isotherm studies was used to analyse the
adsorption isotherms in order to estimate the constants, adsorption
density and adsorption maxima. The experimental results were
fitted to both the Langmuir and the Freundlich isotherms, which are
as follows:
The Langmuir :

x Q 0bC
=
m 1 + bC

(1)

The Freundlich:

x
= KC 1 / n
m

(2)

where:
C is equilibrium concentration (g/l);
x is mass of solute adsorbed (g);
m is mass of adsorbent (g);
b is a constant related to the energy or net enthalpy of adsorption (l /g);
Q0 is mass of adsorbed solute completely required to saturate
a unit mass of adsorbent (g /g);
K is the Freundlich constant indicative of the adsorption
capacity of the adsorbent (l/g); and

Available on website http://www.wrc.org.za

In the adsorption process, fluid flows continuously through a

column of adsorbent or a packed bed, where dynamic adsorption of
the solute occurs (Faust and Aly, 1987). Fixed bed adsorption
columns operated in the downflow mode may perform two functions, namely adsorption and filtration. Breakthrough or exhaustion curves were constructed from the column data by plotting the
ratio of Ce/C0 with the volume of water processed. In the present
studies, the Thomas model was used to study the dynamic behaviour of the column. The model is shown below (Reynolds and
Richards, 1996):

Ce
=
C0

1
k
1 + exp[ ( q0 m C 0V )]
Q

(3)

where:
Ce is effluent adsorbate concentration (g/l);
C0 is influent adsorbate concentration (g/l);
k is Thomas rate constant (ml/min.g);
Q is volumetric flow rate (ml/min);
qo is maximum solid phase concentration (g/g); and
V is throughput volume (ml).
Analytical methods
All samples were acidified (pH < 2) with 0.3% HNO3 (trace metal
grade) and analysed for arsenic using a Varian type SpectrAA600
Zeeman GFAAS equipped with a GTA 100graphite tube atomiser
and programmable sample dispenser (Limit of Detection of the
instrument is 2 g/l). Pyrolytically coated notched partition graphite tubes (Varian Canada Inc., Toronto) were used in the experiments. Argon gas of ultrahigh purity (99.995%; Praxair Products
Inc, Ontario) was used to sheath the atomiser and to purge it
internally. An arsenic hallow cathode lamp (Varian Canada Inc.,
Toronto) was used with emitting wavelength of 193.7 nm with a slit
width of 0.5 nm. Palladium solution (1500 g/l) + magnesium
nitrate (1 000 g/l) solution was used as matrix modifier for
calibration. Fresh modifier solution was prepared for calibrating
the instrument. An external reference standard from the National
Water Research Institute (NWRI), Environment Canada, Ontario
was used to verify the calibration.
The iron content of the GFH was determined by acid digestion
using the procedure described in AWWARF (1993). One gram of
GFH was added to 50 ml of 10% HNO3 in a beaker and the solution
was heated on a hot plate to boiling point. After 2 h, the iron oxide
in the medium was completely dissolved and the acid solution
turned yellow. At this point, digestion was discontinued, the
solution was made up to 1 l with distilled deionised water, filtered
through 0.45 m filter, and the iron content determined by a
FerroVer method using Hach instrument (model DR/850, Hach
Company, Loveland, CO, USA). The BET surface area of the GFH
was determined using Flowsorb 2300 (Micromeritics Instrument
Corporation, Georgia, USA). Single point surface area measurements were employed to determine the surface area of the samples.

Results and discussion

Batch kinetic studies
Figure 1 shows that arsenic (both As(III) and As(V)) was successfully removed from the solution with time. More than 80% arsenic

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

163

Concentration of As(V) in solution

C ( g/l)

120
(b)

100

pH 5
pH 6
pH 7.6
pH 8.5

80
60
40
5 g/L

20
0
0

Time, t (h)
Figure 1
Arsenic concentration remaining with time using GFH a) arsenite and b) arsenate

removal was achieved after 3 h of contact

time with less than 5 g/l achieved at pH
levels of 6 and 7.6 for both As(III) and
As(V) at an equilibrium time of six hours.
The most efficient As(III) adsorption on to
GFH occurred at a pH of 7.6, with 68% of
As(III) removed within 1 h and 97% removed at the equilibrium time of 6 h. The
kinetic study data were fitted to Ho pseudo
second-order reaction rate model (1996) to
describe the adsorption kinetics. Non-linear regression was performed with Statistica
for Windows (release 5.1) software (Statsoft
Inc., 1997) using Quasi-Newton method
(start values were 0.1 for all parameters;
initial step size for all parameters was 0.5;
maximum number of iterations was 50).
The rate constant for adsorption k for both
As(III) and As(V) was estimated as 0.003
and 0.002 g GFH/h.g As, respectively.
Arsenite adsorption was similar to results reported in previous studies (Pierce
and Moore, 1980; Wilkie and Hering, 1996)
where the adsorption of As(III) on to amorphous ferric hydroxide increased as the pH
increased, with maximum adsorption occurring at a pH of 7. A maximum removal
(96%) of As(V) occurred at a pH of 7.6 at an
equilibrium time of 6 h. More than 90%
removal of As(V) was observed at all the pH
levels studied. These results were similar to
the results of earlier studies (Hsia et al.,
1994; Wilkie and Hering, 1996). In studies
with hydrous ferric hydroxide (HFO), close
to 100% As(V) removal was observed in the
pH range of 4 to 8 by Hsia et al. (1994) and
Wilkie and Hering (1996).
Batch isotherm studies

Figure 2
Isotherm plot for arsenic adsorption onto GFH a) the Langmuir and
b) the Freundlich

164

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

Results of the isotherm studies were fitted

into both the Freundlich and the Langmuir
isotherms. Non linear regression was performed with Statistica for Windows (Release 5.1) software (Statsoft Inc., 1997)
using the Quasi-Newton method (start values, initial step size and number of iterations were as before). The estimated adsorption densities by the Langmuir and the
Freundlich models with the concentrations
of As(III) and As(V) remaining in solution
are shown in Fig. 2. All the isotherms were
convex curves.
The separation factor R (Hall et al.,
1996) estimated from the Langmuir constant for As(V) and As(III) was 0.33 and
0.14 (0<R<1), respectively. This suggested
that arsenic adsorption can be modelled by
the Langmuir isotherm. The K parameters
estimated from the Freundlich isotherm for
As(V) and As(III) were 10.3 and 18 (L/g),
respectively. High levels of K suggested
that adsorption capacity of the adsorbent is

Available on website http://www.wrc.org.za

high (AlDuri, 1996). Low 1/n values

(<1) from the Freundlich isotherm suggested that any large change in the equilibrium concentration of arsenic would
not result in a marked change in the
amount of arsenic sorbed by GFH. The
correlation coefficient (r) for all the
isotherms ranged between 0.97 and 0.99,
representing an excellent fit of the observed data. The t-test values showed
that the coefficients were significant at
the 95% confidence level for all the
model equations.
In the case of As(V) removal, the
adsorption maximum and the adsorption density at a residual concentration
of 5 g As/l were estimated through the
Langmuir model and were 159 and 32
g As/g GFH, respectively. Acid digestion results showed that 1 g GFH contained 269 mg Fe. By expressing the
mass of GFH in terms of Fe content and
the concentration of As(V) in mol/l,
the adsorption maximum was estimated
at 17.6 mol As/g Fe (1 mmol As/mol
Fe). The adsorption density was 3.4
mol/g Fe (0.2 mmol As/mol Fe). The
estimated adsorption density was lower
than the value reported by Driehaus et
al. (1998) and Fuller et al. (1993). The
difference could be attributed to different experimental conditions discussed
below. In the batch studies for As(V)
removal from the synthetic solution at
pH 7, Driehaus et al. (1998) reported
that at a residual As(V) concentration of
10 g/l, the adsorption density on GFH
was 1 mmol As/g Fe, at high initial
As(V) concentration. They also reported
that arsenate adsorption on freshly prepared ferric hydroxide was higher than
Figure 3
the adsorption on GFH. It is expected
Arsenic removal from Kelliher water using GFH a) As species remaining in solution and
that at a high initial As concentration or
b) isotherm plot
at a high As/Fe ratio, the adsorption
maximum and the adsorption density
will be also high. High adsorption density was reported by Fuller
In the isotherm studies with natural water, the initial total As,
et al. (1993) in their adsorption studies, where the initial As/Fe ratio
soluble As, and particulate As in the raw water were determined as
was 0.12. In the present studies the initial As/Fe ratio was main177.3, 169.8, and 7.5 g/l, respectively. At the end of isotherm
tained between 0.025 and 0.0025. Further, the surface area of GFH
studies, samples were collected and speciated to find soluble As,
(112 m2/g) estimated in this study was smaller than the value
As(III) and As(V). The speciation of raw water showed the domireported by Fuller et al. (1993).
nance of As(III) species over As(V). The As(V)/As(III) ratio in the
In the studies on As(III) removal, the adsorption maximum
raw water was 0.45. The results showed that the minimum concenand the adsorption density at 5 g/l were estimated at 112 and 31
trations of soluble As, As(III) and As(V) achieved were 4.8, 3.3,
g As/g GFH, respectively from the Langmuir model. In terms of
and 1.6 g/l, respectively. A plot of the concentration of As
available Fe content in GFH, the adsorption maximum and adsorpremaining in solution vs. the mass of GFH is shown in Fig. 3 (a).
tion density at 5 g/l were 11 mol As/g Fe (0.7 mmol As/mol Fe)
The results were fitted into the Freundlich and the Langmuir
and 1.7 mol/g Fe (0.1 mmol As/mol Fe), respectively. The
models. The model curves are shown in Fig. 3 (b). As before, the
estimated adsorption density was smaller compared to the value
isotherm had convex curves. The estimation of isotherm paramreported by Wilkie and Hering (1996). However, the percent of
eters such as K value (18.5) and 1/n value (0.6) confirmed that the
As(III) removal observed in the present studies was higher than the
adsorption capacity of GFH was high. A statistical t-test confirmed
value reported by Wilkie and Hering (1996). The initial As/Fe ratio
that the coefficients were significant at the 95% confidence level.
maintained in their studies was high compared to the present
The adsorption density at a residual concentration of 5 g As/l was
studies.
estimated at 24 g As/g GFH (2.6 mol As/g Fe or 0.15 mmol As/

Available on website http://www.wrc.org.za

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

165

TABLE 2
Isotherm parameters for arsenic removal using GFH
Isotherm parameters

Arsenic species
As(III)

As(V)

Soluble As

Adsorption maxima,g As/g GFH

(mmol As/mol Fe)

112 (0.67)

159 (1)

141 (0.8)

Adsorption density,g As/g GFH

(mmol As/mol Fe)

31 (0.1)

32 (0.2)

24 (0.15)

The Freundlich isotherm

Regression coefficient, r
K value
n value

0.99*
18
2.3

0.97*
10.3
1.5

0.97*
18.5
1.65

The Langmuir isotherm

Regression coefficient, r
Separation factor, R

0.97*
0.14

0.98*
0.33

0.95*
0.22

* coefficients are significant at the 95% confidence level.

As concentration remaining
in solution, C ( g/l)

120
pH - 7.6
Water - tap water
Initial As concentration
- 500 g/L
Flow rate - 21.5mL/min
(5m/h)
EBCT - 2 min

100
80
60

(a)
As(III)-1
As(III)-2
As(III)-3
As(III)-4
As(III)-5
As(V)-1
As(V)-2

40

As(V)-3

5 g/L

As(V)-4

20

As(V)-5

0
0

500

1000

1500

2000

2500

Bed volume
1.2
As(III) observed

(b)

As(III) estimated
As(V) observed

Ce/Co

0.8

As(V) estimated
pH - 7.6
Water - tap water
Initial As concentration 500 g/L
Flow rate - 21.5mL/min
(5m/h)
EBCT - 2 min

0.6
0.4
0.2
0
0

500

1000

1500

2000

2500

3000

3500

Bed volume
Figure 4
Arsenic removal from tap water in column studies a) As concentration remaining in
solution with throughput volume and b) estimated values of Ce/Co by the Thomas
model

166

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

mol Fe) from the Langmuir model.

The results of the batch studies are
summarised in Table 2.
Column studies
Figure 4 (a) shows the concentration
of arsenic remaining in solution vs.
the bed volumes achieved for all the
cycles up to an effluent arsenic level
of 100 g/l. In the first cycle of column tests for As(V) removal, the
column continued to remove As(V)
to a level of less than 5 g/l for a
contact time of 38 h. The bed volumes achieved up to 5g/l were
1 140. Similarly, the bed volumes
achieved for As(III) removal in the
first cycle were 1 260. The bed volumes achieved in the remaining cycles decreased. This could be ascribed
to the observed loss of GFH along
with the effluent during the backwash operations. The bed volumes
achieved for both As(V) and As(III)
up to 10 g/l were 3 240 and 3 000,
respectively. The bed volumes
achieved for As(V) removal in the
present study were high compared to
the values reported in the studies using ion-exchange resin (Clifford et
al., 1999), and lower than the values
reported by Driehaus et al. (1998). In
the laboratory studies using an ionexchange resin, Cilfford et al. (1999)
achieved bed volumes of 400 to 800
up to a breakthrough arsenic concentration of 2 g/l, where the influent

Available on website http://www.wrc.org.za

540
0.27

0.03
(1.7)
0.05
0.91*
1523
0.15
540
0.28
720
0.36

4
500
21.5
60
43
2
54

1020
0.51

3
500
21.5
60
43
2
67

0.04
(2.2)
k value from the Thomas model (ml/ming)
0.22
Regression coefficient, r
0.93*
Bed volumes achieved up to 5 g/l of As
1260
Arsenic removal capacity of the media to 5 g/l 0.63
(mg/cm3)

Column cycles
Initial As concentration (g/l)
Flow rate (ml/min)
Mass of GFH (g)
Volume of the media (ml)
EBCT (min)
Total Throughput volume (l)
Maximum solid phase concentration, qo
(mg/g) from the Thomas model
qo mmol As/g Fe (mmol As/ mol Fe)

1
500
21.5
60
43
2
137
0.88

2
500
21.5
60
43
2
85

1020
0.51

3
500
21.5
60
43
2
70

780
0.39

4
500
21.5
60
43
2
54

540
0.27

5
500
21.5
60
43
2
57

480
0.24

0.09
(5)
0.22
0.92*
1140
0.56

1
500
21.5
60
43
2
126
0.82

2
500
21.5
60
43
2
85

As(V)
As(III)
Arsenic species

Available on website http://www.wrc.org.za

* coefficients are significant at the 95% confidence level.

1
169.3
21.5
40
30.5
1.4
93
0.3
5
500
21.5
60
43
2
54

Soluble
As

Kelliher
water
Tap water
Water

TABLE 3
Summary of the results from column studies using GFH

had an arsenic concentration of 21 g/l. The influent As(V)

concentration maintained in the pilot studies (two in-line columns)
by Driehaus et al. (1998) ranged between 15 and 180 g/l. A low
initial concentration and passage of arsenic through two columns
packed with GFH might be the reason for high bed volumes.
Similarly, it is expected that high bed volumes could be achieved
using GFH, while operating under low initial As concentration.
The results from the first cycle for both As(III) and As(V) were
fitted into the Thomas model. Non-linear regression was per-

formed with Statistica for Windows (release 5.1)

software (Statsoft Inc., 1997) using the HookeJeeves method (start values were 0.1 for all parameters; an initial step size of 2 for all parameters;
maximum number of iterations at 50). The model
values of Ce/C0 are shown in Fig. 4 (b). The
adsorptive capacity or maximum solid phase concentration q0 estimated from the Thomas model
for As(III) and As(V) was 0.88 and 0.82 mg/g GFH,
respectively. In terms of Fe content q0 for As(III)
and As(V) was estimated at 0.04 mmol As/g Fe (2.2
mmol As/mol Fe) and 0.09 mmol As/g Fe (5 mmol
As/mol Fe), respectively. The As(V) adsorptive
capacity of GFH (0.82 mg/g) estimated in the
present study was similar to the value (0.8 g/kg
GFH) achieved by Driehaus et al., (1998) in one of
the pilot facilities (W) located in Germany.
The arsenic removal capacities to achieve levels
of less than 5 g/l by GFH in the first cycle of
column tests for As(III) and As(V) were 0.63 (0.39)
and 0.56 mg/cm3 GFH (0.35 mg/g GFH), respectively. The arsenic removal capacity was calculated
by the difference between the applied arsenic loading and the amount removed by GFH, divided by
the volume of GFH used in the column tests. These
values were higher than the values reported by
Simms and Azizian (1997) using activated alumina.
In a pilot plant study for arsenic removal from
natural water using activated alumina (AA) at a pH
of 7.5, they reported that the arsenic removal capacity of AA to 10 g/l varied between 0.19 and 0.35
g As/kg AA at different empty bed contact times.
The results of the present column studies are summarised in Table 3. Data pertaining to arsenic
recovery through backwash operations at the end of
column operations are shown in Table 4. The
average As(V) recovery efficiency (82.5%) for the
first four cycles was slightly higher than the recovery efficiency (81.5%) for As(III).
In the column studies with Kelliher water using
GFH, the results (Fig. 5 (a)) showed the dominance
of As(V) species after 24 h of column operation
because of oxidation of As(III) to As(V). Similar to
the column studies with Regina tap water, high bed
volumes (1 523) were achieved up to 5 g/l of
soluble As. The column data were fitted to the
Thomas model. The model (Fig. 5 (b)) fitted well
with the observed data. The q0 value was estimated at 0.28 mg/g GFH (0.03 mmol As/g Fe or 1.7
mmol As/mol Fe).
Application to practice

Despite the fact that several technologies have been

proven to be successful in the removal of arsenic from drinking
water at laboratory and pilot-scale studies, the practical applicability of a number of such systems to small communities has not been
fully tested and exploited. In addition to a high arsenic removal
efficiency, the system should be simple, economically viable and
acceptable to the community. Although coagulation-assisted
microfiltration and ion-exchange systems may be suitable for large
communities, systems based on adsorption/filtration processes are
appropriate and advantageous to small communities, especially in

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

167

160
Soluble As
As(III)
As(V)

Arsenic remaining
in solution, C ( g/l)

140
120
100

(a)

Water - natural water

Initial As concentration 177.3 g/L
Flow rate - 21.5mL/min
(5m/h)
EBCT - 1.42 min

80
60
40

5 g/L

20
0
0

500

1000

1500

2000

2500

3000

3500

Bed volume

1
0.9

Observed values

0.8

estimated values

Ce/Co

0.7

(b)

Figure 5
Plot showing a) the arsenic
concentration remaining in
Kelliher water in the column
studies using GFH and b) the
estimated values of Ce/Co by
the Thomas model

Water - natural water

Initial As concentration 177.3 g/L
Flow rate - 21.5mL/min
(5m/h)
EBCT - 1.42 min

0.6
0.5
0.4
0.3
0.2
0.1
0
0

500

1000

1500

2000

2500

3000

3500

Bed volume
developing countries such as Bangladesh. This is due to factors
such as simplicity, ease of construction, operation and maintenance. Currently, simple household purification systems such as
bucket filtration systems containing sand and iron filings are used
in a number of arsenic-affected areas in Bangladesh as short-term
measures. However, a permanent solution has to be developed to
provide appropriate treatment facilities to supply the drinking
water needs of people in small communities. GFH based filtration
systems offer a competitive choice amongst the treatment systems
available for arsenic removal in small water facilities.

Conclusions
The following conclusions were made based on this study:

168

GFH is effective in reducing both As(III) and As(V) to a level

less than 5 g/l in drinking water.
The kinetic study showed that less than 5 g/l of As could be
achieved at the pH levels of 6 and 7.6, with highest arsenic
removal observed at pH 7.6 (the normal pH of the Regina tap
water). The equilibrium time was found to be 6 h.
In the isotherm studies, the observed data fitted well with both
the Freundlich and the Langmuir models and the model equa-

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

tions were found to be statistically significant at the 95%

confidence level. In the studies with Regina tap water the
adsorption densities estimated from the Langmuir model for
both As(V) and As(III) at a residual concentration of 5 g/l
were 3.4 (0.2) and 1.7 mol As/g Fe (0.1 mmol As/mol Fe),
respectively.
The speciation of natural water containing arsenic showed that
the As(V)/As(III) ratio in the raw water was 0.45. However,
conversion of As(III) to As(V) species was observed in the
column studies.
The results of five cycles of column tests with Regina tap water
showed that the bed volumes and arsenic removal capacity
were high, which suggested that GFH-based filtration systems
could be effectively used in small water utilities.

Acknowledgements
The authors acknowledge major support from Health Canada,
Ottawa for this study. The first author expresses sincere thanks to
the officials of Kelliher Water Treatment Facility (Kelliher, Saskatchewan, Canada) for their assistance in the collection of water
samples. The authors thank the reviewers for their suggestions.

Available on website http://www.wrc.org.za

TABLE 4
Arsenic recovered from the column studies after backwash operations
using GFH
Adsorbent

As
species

Column
test
cycles

Total
As load
to the
column
(mg)

As
removed
in the
column
tests
(mg)

As
recovered
by
regeneration
and back
wash (mg)

As
recovery
efficiency
(%)

GFH

As(III)

1
2
3
4
1
2
3
4

68
43
35
27
63
43
34
27

66
42
34
26
61
42
33
26

50
35
30
22
51
35
27
22

75
82
87
82
84
82
81
83

As(V)

References
ALDURI B (1996) Adsorption modeling and mass transfer. In: G McKay
(ed.) Use of Adsorbents for the Removal of Pollutants from
Wastewaters, 138, CRC press, New York.
AWWARF (AWWA Research Foundation) (1993) NOM Adsorption onto
Iron Oxide-Coated Sand. AWWA, Denver, Colorodo.
CHADA (2000) High arsenic in sedimentary aquifers in parts of the Ganga
basin, West Bengal, India. In: P Bhattacharya and AH Welch (eds.)
Proc. of the 31st Int. Geol. Congr., Arsenic in Groundwater of
Sedimentary Aquifers, Rio de Janeiro, Brazil, 33-35.
CHENG RC, LIANG S, WANG HC and BEUHLER MD (1994) Enhanced
coagulation for arsenic removal. J. AWWA 86 (9) 79-90.
CHOWDHURY UK, BISWAS BK, DHAR RK, SAMANTA G, MANDAL
BK, CHOWDHURY TR, CHAKRABORTI D, KABIR S and ROY S
(1999) Groundwater arsenic contamination and suffering of people in
Bangladesh. In: WR Chappell, CO Abernathy and RL Calderon (eds.)
Arsenic Exposure and Health Effects, 165-182. Elsevier, New York.
CLIFFORD DA, GHURYE G and TRIPP AR (1999) Development of
anion exchange process for arsenic removal from drinking water. In:
WR Chappell, CO Abernathy and RL Calderon (eds.) Arsenic Exposure and Health Effects, 379-388. Elsevier, New York.
DRIEHAUS W, JEKEL M and HILDEBRANDT U (1998) Granular ferric
hydroxide-a new adsorbent for the removal of arsenic from natural
water. J. Water Supply: Res. and Technol.-Aqua 47 30-35.
FAUST SD and ALY OM (1987) Adsorption Processes for Water Treatment. Butterworths, Boston. 134.
FULLER CC, DAVIS JA and WAYCHUNAS GA (1993) Surface chemistry of ferrihydrite: Part 2. Kinetics of arsenate adsorption and
coprecipitation. Geochem. et Comochim. Acta 57 2271-2282.
HALL KR, EAGELETON LC, ACRIVOS A and VERMEULEN T (1966)
Pore- and solid diffusion kinetics in fixed bed adsorption under
constant pattern conditions. Ind. Eng. Chem. Fundamental 5 212-223.
HO YS, WASE DAJ and FORSTER CF (1996) Kinetic studies of competitive heavy metal adsorption by sphagnum moss peat. Environ. Technol.
17 71-77.
HSIA TH, LO SL, LIN CF and LEE DY (1994) Characterization of arsenate
adsorption on hydrous iron oxide using chemical and physical methods. Colloids Surfaces A: Physicochem. Eng. Aspects 85 1-7.
JOSHI A and CHAUDHURI M (1996) Removal of arsenic from ground
water by iron oxide coated sand. J. Environ. Eng. 122 (8) 769-771.
KARIM MM (2000) Arsenic in groundwater and health problems in
Bangladesh. Water Res. 34 (1) 304-310.
KORTE NE and FERNANDO Q (1991) A review of arsenic (III) in
groundwater. Crit. Rev. Environ. Control 21 (1) 1-39.
MA HZ, XIA YJ, WU KG, SUN TZ and MUMFORD JL (1999) Human
exposure to arsenic and health effects in Bayingnormen, Inner Mon-

Available on website http://www.wrc.org.za

golia. In: WR Chappell, CO Abernathy and RL Calderon (eds.)

Arsenic Exposure and Health Effects, 127-132. Elsevier, New York.
MAZUMDER DNG, HAQUE R, GHOSH N, DE BK, SANTRA A,
CHAKRABORTY D and SMITH AH (1998) Arsenic levels in drinking water and the prevalence of skin lesions in West Bengal, India. Int.
J. Epidemiol. 27 871-877.
McARTHUR JM, RAVENSCROFT P, SAFIULLAH S and THIRLWALL
MF (2001) Arsenic in groundwater: Testing pollution mechanism for
sedimentary aquifers in Bangladesh. Water Resour. Res. 37 (1) 109117.
NATIONAL RESEARCH COUNCIL (1999) Arsenic in Drinking Water.
National Academy Press, Washington, DC, ISBN 0-309-06333-7.
NICKSON RT, McARTHUR JM, RAVENSCROFT P, BURGESS WG
and AHMED K M (2000) Mechanism of arsenic release to groundwater,
Bangladesh and West Bengal. Appl. Geochem. 15 403-413.
PIERCE LM and MOORE BC (1980) Adsorption of arsenite on amorphous iron hydroxide from dilute aqueous solution. Environ. Sci.
Technol. 14 (2) 214-216.
PIERCE LM and MOORE BC (1982) Adsorption of arsenite and arsenate
on amorphous iron hydroxide. Water Res. 16 1247-1253.
RAVEN KP, JAIN A and LOEPPERT RH (1998) Arsenite and arsenate
adsorption on ferrihydrite: kinetics, equilibrium, and adsorption envelopes. Environ. Sci. Technol. 32 344-349.
REYNOLDS TM and RICHARDS PA (1996) Unit Operations and
Processes in Environmental Engineering. PWS publishing company,
Boston, ISBN 0-534-94884-7.
SANCHA AM (1999) Full scale application of coagulation processes for
arsenic removal in Chile: A successful case study. In: WR Chappell,
CO Abernathy and RL Calderon (eds.) Arsenic Exposure and Health
Effects, 373-378. Elsevier, New York.
SCOTT NK, GREEN FJ, DO DH and MCLEAN JS (1995) Arsenic
removal by coagulation. J. AWWA 87 (4) 114-126.
SIMMS J and AZIZIAN F (1997) Pilot-plant trials on the removal of arsenic
from potable water using activated alumina. In: Proc. of the AWWA
Water Quality Technol. Conf., Denver, Colorado, AMICUS NO
20438745.
SMITH AH, GOYCOLEA M, HAQUE R and BIGGS ML (1998) Marked
increase in bladder and lung cancer mortality in a region of Northern
Chile due to arsenic in drinking water. Am. J. Epidemiol. 147 (7) 660669.
STATISTICA FOR WINDOWS. RELEASE 5.1. (1997) Statsoft Inc.,
Tulsa, USA, www.statsoftinc.com.
SUBRAMANIAN KS and KOSNET MJ (1998) Human exposures to
arsenic from consumption of well water in West Bengal, India. Int. J.
Occup. Environ. Health. 4 (4) 217-230.
THIRUNAVUKKARASU OS, VIRARAGHAVAN T and SUBRAMANIAN KS (2001) Removal of arsenic in drinking water by iron

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

169

oxide-coated sand and ferrihydrite - Batch studies. Water Qual. Res.

J. Can. 36 (1) 55-70.
TSENG WP, CHU HM, HOW SW, FONG JM, LIN CS and YEH S (1968)
Prevalence of skin cancer in an endemic area of chronic arsenicism in
Taiwan. J. Natl. Cancer Inst. 40 (3) 453-463.
USEPA (2000) Proposed Revision to Arsenic Drinking Water Standard.
http://www.epa.gov/safewater/arsenic.html (June 22, 2000).
USEPA (1999) Technologies and Costs for Removal of Arsenic from
Drinking Water. Draft report, EPA-815-R-00-012, Washington DC.
VIRARAGHAVAN T, SUBRAMANIAN KS and SWAMINATHAN TV
(1994) Drinking water without arsenic: A review of treatment technologies. Environ. Systems Rev. 37 1-35.

170

ISSN 0378-4738 = Water SA Vol. 29 No. 2 April 2003

VIRARAGHAVAN T, THIRUNAVUKKARASU OS and SUBRAMANIAN KS (2000) Iron oxide-coated sand - An excellent adsorbent
for arsenic removal in drinking water. In: MM Sozanski (ed.) Proc. of
the 4th Int. Conf. on Water Supply and Water Quality, Krakow,
Poland, September 2000. 1013-1024.
WARD CB (2000) Arsenic. Water and Wastewater Int. 15 (2) 10-14.
WELCH AH, HELSEL DR, FOCAZIO MJ and WATKINS SA (1999)
Arsenic in groundwater supplies of the United States. In: WR Chappell,
CO Abernathy and RL Calderon (eds.) Arsenic Exposure and Health
Effects, 9-17, Elsevier, New York.
WILKIE JA and HERING JG (1996) Adsorption of arsenic onto hydrous
ferric oxide: Effects of adsorbate/adsorbent ratios and co-occuring
solutes. Colloids Surfaces A: Physicochem. Eng. Aspects 107 97-110.

Available on website http://www.wrc.org.za